Review and Measurements of Benchmark AHB2 Amp

[josh358]

That's often been said but really overrated. Complex or not, the amplifier would see just one value at any moment, one value at a time.. That is a concept that while obvious, I never thought of it until I read one of Elliot Sound's article. Again I am not saying testing with just a single sine wave is all that is needed, just saying the point about music's complex waveform is often overrated. We are actually lucky that we have ASR, Amir at least would measure just about whatever he could think of that his AP can measure, including the 32 tone test even if he might not believe it was important to do.

Transient intermodulation distortion is one case in which transient behavior differs from steady-state behavior and isn't detected by steady-state measurements. Modern amplifiers shouldn't suffer from this, but it's evidence of how a discrepancy can occur. There could be, hypothetically, other mechanisms, e.g., transient behavior in the power supply -- voltage droop, what have you. (I don't remember whether the AHB2 PS is regulated, just that it uses rail switching.)

 

[peng]

I'm fully with you on that. But which discrete values are seen here? And this is exactly the grey area, which is very complex and cannot be answered with a voltage divider assuming a constant voice coil resistance of the loudspeaker and certainly not with transients. And the different transducer concepts will also have a great influence on this.

The short paragraph on '7.4.3 Damping factor' is sufficient to understand how complex the theory can be and I think this one is closer to reality:
Link

Not that complicated at all, not for amplifier considerations. He talked about 20, a single number as not adequate, but people who cited the number 20 was just trying to make a point with shock therapy, sort of. Benchmark's article also made that point but they never said 20 or even 100 would be enough, but when you reach >100, even a single value with the stated condition would seem adequate because then under much worse conditions, the value would still be much higher than 20. If not, I would just avoid that kind of speakers for all intents and purposes.

 

[peng]

Transient intermodulation distortion is one case in which transient behavior differs from steady-state behavior and isn't detected by steady-state measurements. Modern amplifiers shouldn't suffer from this, but it's evidence of how a discrepancy can occur. There could be, hypothetically, other mechanisms, e.g., transient behavior in the power supply -- voltage droop, what have you. (I don't remember whether the AHB2 PS is regulated, just that it uses rail switching.)

Oh yes there are things that are not measured on bench tests for sure, no arguments from me. There had been lots of discussions on TIM but not any more, apparently it is much of a challenge for amp designers to overcome, at least not to the good ones I guess. Even just IMD, if the amp has very low THD, IMD would tend not to be an issue as well according to again, another Elliot Sound article I remember reading and it made sense (logical) to me at the time.

 

[rcstevensonaz]

That's often been said but really overrated. Complex or not, the amplifier would see just one value at any moment, one value at a time.

I thought the issue wasn't the complexity of the waveform at a specific point in time (i.e., the note being played); but rather, the "complexity" of the rapid change in dynamics (i.e., power draw) between the preceding note, the note being played, and the note that follows.

 

[josh358]

Oh yes there are things that are not measured on bench tests for sure, no arguments from me. There had been lots of discussions on TIM but not any more, apparently it is much of a challenge for amp designers to overcome, at least not to the good ones I guess. Even just IMD, if the amp has very low THD, IMD would tend not to be an issue as well according to again, another Elliot Sound article I remember reading and it made sense (logical) to me at the time.

Yep, TIM and SID have been a non-issue for many years now since once designers became aware of it, they were able to design for better open loop behavior.

 

[maxwell3e10]

I am wondering what is a typical DC offset on the output of the amplifier with inputs shorted. Has anyone measured it with a DC voltmeter?

 

[peng]

I am wondering what is a typical DC offset on the output of the amplifier with inputs shorted. Has anyone measured it with a DC voltmeter?

Are you talking about the AHB2 specifically? I don't own one but I bet it would be less than 10 mV, probably close enough to 0.

 

[peng]

I thought the issue wasn't the complexity of the waveform at a specific point in time (i.e., the note being played); but rather, the "complexity" of the rapid change in dynamics (i.e., power draw) between the preceding note, the note being played, and the note that follows.

Okay then that's different but still not an issue unless the amp has very low slew rate. If I remember right, the AHB2's slew rate is 16V/µS as posted on this thread long time ago. The amp is rated 100 W average (8 ohm) so 200 W peak. The math will show that its slew rate is not an issue at all as no music transients in the audio band including the low order harmonics would be too fast for the amp. If it is fast enough for each moment, then there should be no issue with the next (preceding, follow etc. that you mentioned). Do you think that is an issue?

 

[maxwell3e10]

Are you talking about the AHB2 specifically? I don't own one but I bet it would be less than 10 mV, probably close enough to 0.

Yes, I just got one, but the offset seems a little high, 15 and 16 mV in two channels.

 

[peng]

Yes, I just got one, but the offset seems a little high, 15 and 16 mV in two channels.

That surprise me a little but I don't think that is an issue. Did you also measure it without shorting the input but with no input instead? I measured my new Hypex and Purifi amps with and without the inputs connected, and they were higher than my class AB amps including my AVR, but at about 28 mV on one channel and 9 mV on the other, the Hypex amp is still not bad.

 

[maxwell3e10]

That surprise me a little but I don't think that is an issue. Did you also measure it without shorting the input but with no input instead? I measured my new Hypex and Purifi amps with and without the inputs connected, and they were higher than my class AB amps including my AVR, but at about 28 mV on one channel and 9 mV on the other, the Hypex amp is still not bad.

It doesn't matter if the inputs are shorted or open. It has AC coupling, so after input is connected or disconnected there is a little transient, but at the end it settles to the same value. The gain setting also doesn't make a difference.

 

[peng]

It doesn't matter if the inputs are shorted or open. It has AC coupling, so after input is connected or disconnected there is a little transient, but at the end it settles to the same value. The gain setting also doesn't make a difference.

Agreed, that's what I found anyway, same values no matter..

 

[mdunjic]

THX AAA concept sounds very similar to Quad “current dumping” … both are feed-forward based designs, both use highly accurate error correction class A amp stage in combination with brute force class B (or class AB) final output stage … THX may be modern day / improved version of what Quad “current dumping” revolutionized since early 70s

http://www.audiocircuit.com/A-PDF/AA-Brands/Q/Quad-QUA/Current-dumping-audio-amplifier-A-EN2-920-QUA.pdf

http://www.meridian-audio.info/public/quad_405_test_reports%5B2314%5D.pdf

 

[PASOKO]

Good morning all. I'm Pasoko, the guy who posted before about the coil squeal of the AHB2.
I sent it in for repair after that, but the Japanese distributor said there was nothing wrong with it and returned it to me. But later, the manufacturer sent me a request for collection and replacement, and of course I agreed.
As a result, the problem is solved. I'm having an overwhelmingly wonderful audio life.

If your AHB2 is experiencing abnormal coil squeal, it may be defective. Please send it in for repair.

 

[mdunjic]

I'm fully with you on that. But which discrete values are seen here? And this is exactly the grey area, which is very complex and cannot be answered with a voltage divider assuming a constant voice coil resistance of the loudspeaker and certainly not with transients. And the different transducer concepts will also have a great influence on this.

The short paragraph on '7.4.3 Damping factor' is sufficient to understand how complex the theory can be and I think this one is closer to reality:
Link

All analog electronics amplifiers (preamps or power amps) are not seeing discrete signals (like CD, streamers, DACs), but continuously changing analogue signal at their input terminals (rca or xlr) … they don’t work based on sampling their inputs, but their internal blocks and electronics react to continually changing analogue signals.

 

[iwantobelieve]

Has anyone else noticed the figures posted by Paul Miller at HiFi News for dynamic power vs. distortion?

https://www.scvdistribution.co.uk/files/HFN_Mar_Benchmark%20AHB2_Reprint_1_3.pdf

Seems to be quite different to its continuous load performance?

Interesting by contrast with a Purifi implementation -

NAD C 298 Class D Power Amp Lab Report

Lab Report True to tradition, and despite being a Class D amp, NAD has engineered some useful headroom into the C 298 which supports peaks of 265W, 526W and 750W into 8, 4 and 2ohm loads with 450W/1ohm limited by current protection (21.2A). Note how there is no abrupt clipping point but...

www.hifinews.com

(Also, out of interest, the same test on one of the Quad ‘current dumping’ amps -
https://www.hifigear.co.uk/media/uploads/Quad_QSP_HFN.pdf )

 

[Willem]

It does not look like I need to replace my refurbished Quad 606-2 (an earlier version of the 909 and QSP).

 

[iwantobelieve]

It does not look like I need to replace my refurbished Quad 606-2 (an earlier version of the 909 and QSP).

It would be nice if the Quad amps were fully balanced, though. I don’t think even their monoblocks were, even though some of them had balanced connections?

 

[misterdog]

This being one such version, available with balanced input.

https://www.dadaelectronics.eu/uploads/downloads/03_Quad-User-Manuals/Quad-520f-User-Manual.pdf

 

[iwantobelieve]

This being one such version, available with balanced input.

https://www.dadaelectronics.eu/uploads/downloads/03_Quad-User-Manuals/Quad-520f-User-Manual.pdf

I believe the 520f is not fully balanced in its standard form, despite having balanced input connectors? Wasn’t there a balanced input module from Quad to do this, which then technically added an M to the name? IIRC there were apparently very very few of these in existence.